cross-dimensional visual queries for interactive+animated...
TRANSCRIPT
Cross-dimensional Visual Queries for Interactive+AnimatedAnalysis of Movement
Chris WeaverSchool of Computer Science and the Center for Spatial Analysis
University of OklahomaThe North-East Visualization and Analytics Center
Penn State [email protected]
2
• Relative motion patterns in geospatial data• Algorithms for calculating derived motion attributes
• Developed by Patrick Laube (dissertation, etc.)• Visited GeoVISTA Center for Fall 2006
• Adapted as a data query module in Improvise• Can process batch-style or on-demand
• Inputs: entity, location, time• Outputs: velocity, sinuosity, azimuth, others possible
REMO
3
herd movementa visualization of movements of individuals in a herd of radio-tagged caribou
Data Source: Patrick LaubeVisualization Design: Chris Weaver and Patrick Laube
4
• Filtering with conjunctive combination of range filters over1. each raw and derived point dimension2. set containment in brushed group of individuals
• Peripheral views filtered on1. each others' range selections2. the spatial extent of the map itself
• The visualization overall embodies Shneiderman's mantra in a (nearly) symmetric, multi-D manner– A kind of “overview+detail mesh”– Every view is a detail view– Zoom and filter the “map” through interaction in other views
• Conjunctive semantics of interaction forces drill-down in multiple dimensions to follow a single path up/down; no “sideways” queries
• Reduces analytic utility by limiting space of possible questions/queries
What Worked and What Didn’t
5
• Multiple views support selection over sets or ranges of attribute values in multiple raw or derived data columns, across one or more tables.
• Attributes map into appropriate multi-D views, possibly on a many-to-many basis.
• Each view supports binary categorization of values (selected or not) by selection or navigation.
• Users can rapidly toggle highlighting between pairs of views to pose complex focus+context set queries.
• Analysts can form hypotheses and follow chains of evidence by successive selection/deselection and highlighting/unhighlighting of values.
Cross-Highlighting, Approach
6
Cross-Highlighting, Queries
7
health clinic evacuationa visualization of movements of RDIF-carrying health care workers and visitors
Data Source: VAST 2008 Challenge, Evacuation Mini-Challenge (synthetic)Visualization Design: Chris Weaver and Anthony Robinson
8
Demo
Acknowledgments
• Patrick Laube• Alan M. MacEachren, et. al. at GeoVISTA Center and
NEVAC• Collaborators on various related applications, especially
– Deryck Holdsworth (Penn State/GeoVISTA)– David Fyfe (Penn State/GeoVISTA)– Phil Schrodt (U. Kansas/Polisci)
10
Improvise continues to be…… distributed under the GPL… with many example visualizations and source code… and as a Java Web Start (JNLP) application… at http://www.personal.psu.edu/cew15/improvise/
Thanks!